Entity relation extraction method

An entity relationship and relationship technology, applied in the field of information processing, can solve problems such as high labor costs and inability to deal with Internet volume entities and relationship information.

Active Publication Date: 2018-11-02
PEKING UNIV SHENZHEN GRADUATE SCHOOL
View PDF8 Cites 80 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In today's information explosion, traditional supervised relationship extraction cannot cope with the huge and growing entity and relationship information on the Internet due to the high labor cost required to label training samples.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Entity relation extraction method
  • Entity relation extraction method
  • Entity relation extraction method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0073] Please refer to figure 1 , a flowchart of an entity relationship extraction method. The entity relationship extraction methods disclosed in this application include:

[0074] S301. Input the preprocessing information into the word sequence neural network to extract the first type of extraction relationship. The preprocessing information includes several sentences, and the sentences can be obtained directly from text information or voice information. Extracting relationships includes outputting various relationships represented by preprocessing information and the probabilities of the respective relationships.

[0075] Query the entity referred to by each word in the preprocessing information based on the knowledge graph, and convert the preprocessing information into an entity sequence according to the order of the entities in the preprocessing information;

[0076] Perform word segmentation on the preprocessing information to obtain several words; convert the prepro...

Embodiment 2

[0100] Please refer to figure 2 with image 3 , figure 2 It is a flowchart of an entity relationship extraction method, image 3 It is a schematic diagram of the structure of an entity relationship extraction method. The entity relationship extraction method in this embodiment includes:

[0101] S401. Transform the preprocessed information into an entity sequence, specifically including:

[0102] Through n-gram (referring to n words that appear consecutively) text matching, the sequence of appearance of entities in the preprocessing information is linked to the knowledge graph used, while retaining the link of each entity to refer to the candidate entity. Wherein, entity linking (EntityLinking), or entity linking, refers to linking the name appearing in the preprocessing information to the entity it refers to. In natural language, multiple entities may share the same name, that is, names may be ambiguous. For example, the name "Washington" can refer not only to the fir...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an entity relation extraction method. The method comprises the following steps: inputting pre-processed information into a word sequence neural network and an entity sequence neural network, respectively performing relation extraction, thereby two networks to mutually learn through a bidirectional knowledge distillation way, and integrating a relation prediction result of two networks as the final prediction result to output. Since the pre-processed information is input into two different neural networks, two neutral networks are trained at the same time and mutually used as the teacher of the opposite party to perform the adjusting of the neural network parameter; the weighted integrated output are performed on the extraction relations output by two neural networks, two neural networks are used for removing noise data in the training sample in a cooperative way, the respective advantages of two different neural networks are integrated to realize the aims of optimizing and reducing noise.

Description

technical field [0001] The invention relates to the field of information processing, in particular to an entity relationship extraction method. Background technique [0002] Information Extraction refers to the process of extracting information such as entities, events, and relationships from a piece of text, forming structured data and storing it in a database for user query and use. Relation Extraction is the key content of information extraction, which aims to discover the semantic relationship between entities in the real world. In recent years, this technology has been widely used in many machine learning and natural language processing tasks, including the construction and completion of Knowledge Graph (KG), information retrieval, question answering system, etc. [0003] Traditional relation extraction research generally adopts supervised machine learning methods, which regard relation extraction as a classification problem, use manually labeled training data, and tra...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F17/30G06N3/04
CPCG06N3/045
Inventor 雷凯陈道源沈颖
Owner PEKING UNIV SHENZHEN GRADUATE SCHOOL
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap